5
$\begingroup$

Let $f_n$ be a sequence of continuous, differentiable a.e. functions on $[0, 1]$ with

  1. $f_n \to f$ uniformly for some continuous $f$.
  2. $f'_n - g \to 0$ in $L^\infty$ for some measurable $g$,

where we assume that $f'_n - g$ is in $L^\infty$ for all large enough $n$.

Is it true that $f$ is differentiable a.e. with $f' = g$ a.e.?

Note: We do not assume $f'_n, g$ to be in $L^\infty$ a priori, only their difference.

$\endgroup$
0

2 Answers 2

6
$\begingroup$

If the $f_n$ are absolutely continuous, then $f'_n$ would also be the distributional derivative and $g=f'$ follows from the continuity of that. So if there is a counterexample it should probably involve the Cantor-parts of a sequence of BV function converging to something absolutely continuous.

Indeed I think the following should work: Let $f_0:\mathbb{R} \to \mathbb{R}$ be the standard Cantor-staircase, repeated so that $f_0(x+1)=f_0(x)+1$. Define $f_k(x) := \frac{1}{k}f_0(kx)$. Then $f_k$ converges to $f(x)=x$ uniformly. But $f_k'(x) =0$ a.e. for all $k$, so $g(x) = 0 \neq 1 = f'(x)$ almost everywhere.

$\endgroup$
1
  • 1
    $\begingroup$ Neat counterexample! It seems the idea is that the Cantor parts alone can approximate any continuous function in sup norm. $\endgroup$
    – Nate River
    Commented May 23 at 11:10
2
$\begingroup$

Is $f'_n$ assumed to be the distributional derivative of $f_n$ or just the (a.e.) pointwise one? In the latter case, I think a counterexamples is given by considering a piecewise constant approximation of a nowhere differentiable continuous function.

EDIT: Sorry I missed the assumption of continuity for the $f_n$, the counterexample however would still be in place by combining the piecewise linear approximation of the nowhere differentiable function with the answer of mlk in which is shown how to approximate a linear functions with continuous functions with a.e. zero derivative.

$\endgroup$
1
  • 1
    $\begingroup$ Indeed, and I think one can arrange for the limit $f$ to be any continuous function with a generalisation of the construction by @mlk - just a bunch of scaled Cantor staircases. $\endgroup$
    – Nate River
    Commented May 23 at 11:14

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .